20 Phosphorus Transfer From Tropical Terrestrial To Aquatic Systems - Mangroves
Ignacio H. Salcedo1 and Carmen Medeiros2
1) Setor de Radioagronomia, Dept. de
Energia Nuclear, and
2) Departamento de Oceanografia, Universidade Federal
de Pernambuco, 50740-540, Recife (PE), Brasil
Rivers and coastal lagoons are the major source of sediments and phosphorus to shelf waters. Every year, 22.109 Mg of material are carried to the Earth's continental margins by rivers (Garrels and MacKenzie, 1971). Inland and coastal floodplains benefit from the sediment and P transport from uplands into the ocean. Floodplains are characterised by low hydrodynamical energy relative to rivers as the result of shallow depths and vegetation, interfering with the water flow and creating a depositional environment. Both conditions increase the mean residence time of P within the continent, allowing P to enter various biogeochemical cycles and leading to a significant productivity in those areas (Mann, 1982; Alongi et al., 1989).
Coastal floodplains (including deltas) between latitudes of 30ÚN and 30ÚS are predominantly colonised by mangrove, which is adapted to receive fresh and ocean water inputs. Mangroves also exist outside the tropical belt in regions of warm oceanic currents. Hence, the latitudinal distribution of mangroves tends to be broader on eastern than on the western continental margins (Duke, 1992).
Worldwide, mangroves occupy an area of 1.4.105 km2 (Lacerda et al., 1993) more than 50% of them in Asia (Saenger et al., 1993). Mangroves in Africa contribute with 0.33.105 km2 (John and Lawson, 1990; Diop, 1993), those in Australia with 0.12.105 km2 (Bunt, 1992) and those in Latin America and Caribbean with additional 0.4.105 km2 (Lacerda et al., 1993). Marsh vegetation may co-exist with mangrove in tropical areas but is generally limited to a narrow band between the mangroves and the dryland (Carter, 1988).
In the transfer of sediments and P from land through rivers to mangroves, biological and geochemical transformations occur, whose study requires methods from both terrestrial and aquatic sciences. We therefore re-evaluate some of the methodology in the context of tropical terrestrial and aquatic ecosystems.
PHOSPHORUS IN RIVER WATERS
Most of the phosphorus transported by rivers into estuaries originates from surface runoff from soils, weathered rocks and river banks within the watershed, although contributions from subsurface runoff and groundwater tables can be significant. The quantity and quality of P in a river is a weighted average of the relative contribution of the various soils or rocks, and of the land use in the watershed, including diffuse and point source pollution (Ryden et al., 1973; House and Casey, 1989; Frink, 1991; Ramírez, 1991; Sharpley et al., 1992; Caraco, Ch. 14).
DISSOLVED P
The concentration of dissolved PO4 (SRP) in most unpolluted river waters varies between 0.1-1 µM (Meybeck, 1982) although higher values are not uncommon, particularly in rivers draining industrial or agricultural watersheds (Meybeck, 1982; Subramanian and Vaithiyanathan, 1989; Alongi et al., 1992). At the lower end, values may be unreliable because of analytical interference from organic P (Dick and Tabatabai, 1977) and silica (Ciavatta et al., 1990). The normal concentration range found in soil solutions is similar (Fried and Broeshart, 1967), despite the very large differences in solid:liquid (w/v) ratio; 5:1 in a soil with a 20% moisture content as compared to 10-4:1 in a river with a suspended load of 100 mg l-1. Therefore, if the SRP concentration in the receiving water body remains unchanged with the runoff input, the solids eroded by rainstorms will release P at the initial suspension stage, since rainwater normally contains low P concentrations (0.16 µM, Meybeck, 1982).
An example of P release from eroded material can be derived from the data by Sharpley et al. (1992), using the average soil loss (43 kg ha-1 y-1), runoff water (9.2 cm) and SRP and total particulate P (TPP) amounts (122 and 78 g ha-1 y-1, respectively) for six watersheds under native grass cover (4 unfertilised + 2 lightly fertilised). These values yield a suspended load of 47 mg l-1 (found in many rivers), and concentrations of 4.3 µM of soluble and 2.7 µM of particulate P. Hence, 61% of the TPP content was solubilised. The P export value by Sharpley et al. (1992), 200 g ha-1 y-1, is within the range of exports from undisturbed tropical catchments, of 80-460 g ha-1 y-1 (Lesack et al., 1984; Lewis 1986; Saunders and Lewis, 1988).
The relative contributions of P from various parts of a watershed have been observed in a seasonal study in the Amazon River (Devol et al., 1991). During the low water season, mainstream dissolved PO4 concentrations were high (1.1 µM) and relatively constant along a 1,600 km stretch upstream of the estuary (at Obidos), due to a relatively large contribution of runoff from the Andes mountains. In contrast, during the early falling-water season, a continuous decrease in concentration, from 1.0 µM to 0.6 µM was observed along the same stretch (Devol et al., 1991), reflecting a larger water contribution with lower P concentration from the tributaries (0.45 µM) and from floodplain drainage (0.31 µM). This points to the need for a better characterisation of watersheds (relative contribution by runoff and chemical characterisation of the transported material) to improve our understanding of the composition of river loads.
Dissolved organic P (DOP) can contribute significantly to the total dissolved P load in rivers. Devol et al. (1991), found an average DOP concentration of 0.4 µM in the Amazon River and, unlike the inorganic fraction, DOP concentrations were relatively uniform along a stretch of 1600 km. DOP values reported by Ramírez (1991) for Venezuelan rivers draining into the Caribbean Sea were 0.13 µM, lower than the 0.35 µM for the Orinoco river (Lewis and Saunders, 1989) and than the world average, 0.48 µM (Meybeck, 1982). The weight ratio of dissolved organic C: dissolved organic P (DOC:DOP) in the rivers of the Caribbean Sea basin was 500:1 (Ramírez, 1991), in the same range of those he derived from data by Meybeck (1982), 367:1, and from Lewis and Saunders (1989), 404:1.
PARTICULATE P
Eroded material usually is enriched in fine fractions, silt and clay, which are the most abundant suspended size fractions in rivers, and which contain most of the total P of soils (Tiessen et al., 1983) and sediments (Salomons and Gerritse, 1981). Concentrations of total particulate P (TPP) reported for the Amazon river, tributaries and floodplains are 7.9, 3.0 and 1.6 µM, respectively, with the fine particulate fraction (<63 µm) being more enriched in P, 28 µmol g-1, than the coarser one, 13 µmol g-1 (Devol et al., 1991). The contribution of organic P in the particulate P pool was small: 0.5 µM in the mainstream, 0.6 µM in the tributaries and 0.25 µM in the floodplains. Similar values of TPP, 16 µmol g-1, where found in Venezuelan rivers draining into the Caribbean Sea (Ramirez, 1991). Both, the Amazon and Venezuelan values of TPP are smaller than the average for Indian rivers, 35 µmol g-1 (Subramanian and Vaithiyanathan, 1989), the Zaire River, 68 µmol g-1 (Sholkovitz et al., 1978) and the world average, 37 µmol g-1 (Meybeck, 1982).
CHEMICAL FORMS OF PHOSPHORUS IN SOILS AND SEDIMENTS
For an understanding of the biogeochemical processes undergone by P in rivers and estuarine waters, particulate P has to be characterised by: (1) its chemical composition; (2) the proportion of total P in fast equilibrium with soluble P (at a time scale close to that of the water in the estuary); and (3) the mechanisms responsible for P transfer between solid and solution.
characterization of particulate P
Since the original study by Chang and Jackson (1957), much work has been done to fractionate P into its main chemical forms (associated with Ca, Fe and Al), on both soils (Williams et al., 1967, Petersen and Corey, 1966; Hedley et al., 1982) and sediments (Kurmies, 1972; Hieltjes and Lijklema, 1980; Barbanti and Sighinolfi, 1988; Golterman and Booman, 1988; Ruttenberg, 1992). One recent sequential extraction scheme for marine sediments (Ruttenberg, 1992) describes five P fractions:
• loosely bound (MgCl2 extractable);
• Fe-bound (citrate-dithionite-bicarbonate, CDB);
• authigenic apatite (formed in situ) + calcite-P + biogenic apatite (acetate buffer at pH 4);
• detrital apatite + other inorganic P (1M HCl);
• organic P (ashing at 550Ú+ 1M HCl).
Such fractionations are rarely as specific as originally proposed (Sadler, 1973), in spite of the extensive use of reference materials to evaluate the efficiency of the extractions steps (Five, 1962; Pratt and Lindsay, 1967; Ruttenberg, 1992). In the CDB extraction of Fe bound P, recoveries of added P varied between 82 and 89% (Ruttenberg, 1992). The following MgCl2 washing did not recover the unextracted P suggesting that part of the CDB-solubilised P was readsorbed. In more complex matrix phases, Hamad et al. (1992) showed that the maximum P adsorption capacity of Sudanese soils was reduced only 20%, in average, after removal of Fe with a CDB extraction. The unextracted P could be released by the following extraction (authigenic Ca-P), thus underestimating Fe-P and overestimating authigenic Ca-P.
The use of CDB to extract specifically Fe-P (Ruttenberg, 1992) needs reexamining since Fe and Al are closely associated. X-ray microprobe scans in ferruginous nodules from Ghana and Brazil combined with radioisotope techniques, showed high adsorption of 32P in Al-enriched surface coatings within the Fe nodules (Tiessen el al., 1991). Aluminum can substitute for Fe in crystalline Fe minerals (Ainsworth et al., 1985; Schwertmann, 1991). Due to the close association of Fe and Al, the CDB extraction of soils solubilises significant amounts of both elements (Singh and Gilkes, 1991; Salcedo et al., 1991a). The analysis of Fe and Al, in addition to P, in the CDB extracts from sediments could indicate the relative contribution of these P-retaining materials.
Berner and Rao (1994) used a variation of the P fractionation method proposed by Ruttenberg (1992) to analyze suspended and bottom sediments of the Amazon River. Of the total P in the suspended sediment, 22.9 µmol g-1, or 33%, was organic and 77% was inorganic P. The inorganic P was equally divided between Fe (+Al) and Ca-P forms. Of the Ca-P, 59% was authigenic and 40% detrital apatite. The large Ca-P fraction can probably be attributed to Andean material (Gibbs, 1967; Stallard and Edmond, 1983; Forsberg et al., 1988; Devol et al., 1991), since soils in the Amazon basin are normally described as highly weathered and leached (Sanchez et al., 1991). Highly weathered soils normally contain very low levels of Ca-P (<1 µmol g-1) (Tiessen et al., 1992; Araújo et al., 1993). Readsorption of P dissolved by the CDB extraction could have overestimated authigenic Ca-P at the expense of the Fe-P fraction in Berner and Rao's (1994) results.
Many of these uncertainties may be clarified once extraction methods are combined with electron dispersive X-ray analysis (Berner and Rao, 1994). The presence of mixed P mineral forms, consisting of Fe, Si, Al, Ti and Ca in proportions unrelated to known secondary P minerals in soils, has been recently determined in soils using this methodology (Agbenin and Tiessen, 1994).
In tropical weathered soils organic P can comprise 40 to 70% of the total P content (Sanchez, 1976) and this is reflected in the organic P content of sediments. The dynamic nature of the organic P fraction in soils has been recognised in both, temperate (Stewart and Tiessen, 1987) and tropical soils (Tiessen et al., 1984, 1992). The biologically mediated transformation of organic into inorganic P forms (mineralisation) and vice-versa (immobilisation) occurs simultaneously (Dalal, 1977). Since only small portions of the total organic P pool may be biologically active, the net result of this mineralisation-immobilisation turnover might be difficult to measure (Stewart and Tiessen, 1987). Erosive processes disrupt the soil structure and expose surfaces of previously physically protected organic matter to the action of heterotrophic microorganisms. This shifts the turnover into a net mineralisation, reflected by higher organic P contents in recent suspended sediments, in contrast to lower ones in older bottom sediments of the same river (Sundby et al., 1992; Berner and Rao, 1994). When sediments are deposited in systems with high biological activity, like mangrove swamps, the shift in turnover favours a net immobilisation, resulting in the accumulation of organic P forms, that can account for 80% of the total P content in such sediments (Hesse, 1963).
Extractable P
Extracting a soil or sediment sample once with a single reagent (as opposed to a sequential fractionation of total P) aims at quantifying the proportion of the total P in the sample that might be readily available for biological use. The amount of P extracted has been variously termed extractable (e.g. Hesse, 1961; Boto and Wellington, 1983), exchangeable (Le Mare, 1982), loosely sorbed (Ruttenberg, 1992), labile (Larsen, 1952; Schofield, 1955; Larsen and Sutton, 1963), available (e.g. Chase and Sayles, 1980), bio-available (Sharpley et al., 1992) and biogeochemically available (Howard et al., this vol). The term 'extractable' P together with the definition of the extracting agent, is perhaps the least ambiguous. The extraction of river sediments with P-free seawater has proven to be a good estimate of surface P that could be desorbed in estuaries (Chase and Sayles, 1980; Fox et al., 1986). Another example is the relationship between nitrilotriacetic acid-extractable P and P available to algae (Golterman, 1977). This extractant was also used to extract a fraction of Al and Fe related to P adsorption in soils (Yuan and Lavkulich, 1994).
The term "available" is normally related to a specific P sink and a limited time span. An extraction method rarely will extract an amount of P that is close to that extracted by plants during a growing season (Thomas and Peaslee, 1973), but when the choice of extraction method has been adequate, these two amounts show a good degree of correlation. This has been shown for extractable P from sediments (sodium acetate/acetic acid buffer, pH 4.5, Hesse 1961), which correlated significantly with mangrove growth (Boto and Wellington, 1983).
MECHANISMS CONTROLLING P CONCENTRATION IN RIVER AND ESTUARINE WATERS
It is not clear whether the abiotic regulation of soluble P concentrations in freshwaters is mainly through adsorption-desorption or through precipitation-solubilisation reactions (Frossard et al., this vol). Experimental data representing the partitioning of P between a solid and a liquid phase have been adjusted by various types of adsorption isotherms, both in soils (Barrow, 1987; Hamad et al., 1992) and sediments (Krom and Berner, l980; Jonge and Villerius, l989), and for systems controlled by Fe (Crosby et al., l984), Al (Veith and Sposito, l977a) and Ca (House and Donaldson, 1986). However, it has been recognised that these models are basically curve-fitting techniques, and that no physical-chemical meaning can be ascribed to the parameters in those expressions (Sposito, 1984), since data from precipitation reactions are also well described by, for example, the Langmuir isotherm (Veith and Sposito, 1977b).
Barrow (1987) has proposed a two-step mechanism of initial P adsorption followed by slower diffusion into the solid. For rivers with a favourable Fe:P ratio, Fox (1989) interpreted this two-step mechanism as the formation of a metastable solid-solution of ferric phosphate dissolved in ferric hydroxide. Using a thermodynamic model, he was able to predict the phosphate concentrations in samples from various temperate (Hudson, Mullica and Delaware) and tropical (Amazon, Negro and Sepik) rivers. Similarly, and using data records of water composition covering a 20 year period, Salingar et al. (1993) were also able to explain the P concentration of the Jordan River through a thermodynamic model. The phosphate in solution, in this case, is controlled by the formation of a metastable complex with CaCO3, Ca2(HCO3)2HPO4. Phosphorus in the sediments of the Mississippi river is mostly associated with Ca (Alberts, 1970). Fox et al., (1985) postulated the presence of hydroxyapatite in the soils of the watershed, and suggested that the incongruent solubility of this mineral (Rootare et al., 1962) could control the observed SRP in this river's waters.
Thermodynamic models for rivers (Fox, 1989) are not applicable in high ionic strength estuarine waters (Stumm and Morgan, 1981). Laboratory experiments have shown that bottom sediments from the river and from low salinity sections of the Amazon estuary release soluble P into low-P seawater or into deionised water with increasing salinity. Similarly, bottom sediments from the high salinity region released P into low-P seawater, although with a more complex kinetic pattern (Fox et al., 1986). In these cases, the control of soluble P concentrations in estuarine waters has been attributed to the desorption of P from resuspended sediments (Chase and Sayles, 1980; Fox et al., 1986, Froelich, 1988), rather than to solubilisation reactions. Very low SRP concentrations (0.02 µM) have been attributed to adsoprtion mechanisms (Eyre, 1994).
ESTUARIES AND MANGROVE FORESTS
The P transfer from the land into the ocean is normally mediated by estuaries. The term estuary is here used in its functional concept (Kjerfve, 1989), thus including a diverse and large number of systems such as coastal lagoons, river mouths and deltas. This diversity results from the interaction of geomorphology, freshwater inputs, and tidal and wind forces. Each type of estuary will exhibit a pattern of water circulation and a degree of mixing that will ultimately determine the composition and flux of P to coastal waters.
Three zones can be recognised within an estuary: (1) a tidal river zone, which is a fluvial region lacking ocean salinity but subjected to sea level rise and fall; (2) a mixing zone, characterised by the mixing of water masses and the existence of strong property gradients reaching down to the location of a rive-mouth bar or ebb-tidal delta; and (3) a nearshore turbid zone in the open ocean reaching to the edge of the tidal plume at full ebb tide (coastal boundary layer) (Kjerfve, 1989). Boundaries between the zones continuously change position in response to changing freshwater discharge, astronomical and meteorological forcing.
A zone of maximum turbidity is found in most estuaries in the lower mixing zone (Salinity = 0-8‰). Flocculation of inorganic and organic colloids takes place in this region, contributing to the removal of dissolved inorganic P (Figueres et al., 1978; Sholkovitz et al., 1978; Smith and Longmore, 1980; Morris et al., 1981; Fox et al., 1986). In the regions with higher salinity (>8‰), estuaries can behave as sources (Morris et al., 1981; Kaul and Froelich, 1984; Fox et al., 1986; Jonge and Villerius, 1989; Shengquan, 1993) or sinks of soluble P (Day et al., 1989). This non-conservative behaviour indicates that biogeochemical processes and recycling are acting within the estuary, as opposed to a conservative behaviour, where only dilution is operative (Liss, 1976).
MANGROVE TYPES
Mangrove is the dominant vegetation in 75% of tropical estuaries (Boto and Bunt, 1981). Lugo and Snedaker (1974) identified 6 functional types of mangrove forests (riverine, basin, fringe, scrub, overwash and hammock) for Florida, depending mainly on physiognomic characteristics. Equivalent systems can also be identified, but not as easily, in the old world (Woodroffe, 1992). The mangrove types based on the geomorphological setting and associated delta type are: river-dominated, tidal-dominated, wave-dominated barrier lagoon, wave+river-dominated, drowned bedrock valleys, and carbonate settings (Thom, 1982, 1984).
Functional and geomorphological classifications are interrelated, as they respond to a degree of openness/closeness of the wetlands and to a given balance among riverine, astronomical (tides) and climate-related forcing (rainfall, wind stress, etc.). Mangrove production, material cycling and exports vary with the forest structure and setting. A comparative study including 50 freshwater and saltwater wetland forests was done by Lugo et al. (1988). Observations from these and other authors are summarised in Table 1. Comparisons are only qualitative since available data are still scarce and there is a large variability among systems and methodology for data acquisition.
Table 1. Comparison of characteristic parameters for riverine, fringe, basin and scrub forest mangroves.
Forest type Parameter Riverine Fringe Basin Scrub Freshwater turnover High Moderate Moderate Low Soil salinity Low Moderate Moderate High P inflow High Moderate Moderate Low P outwelling High Moderate Low Low Hydrologic energy from wave+tide+runoff Moderate High Low Low Water flow Seaward along main channel Perpenden-dicular to channel No preferential direction Almost no flow Internal recycling Low Moderate Moderate High H2S sediments Low Moderate Moderate High
Phosphorus transfers through litterfall are higher for systems that are subjected to greater tidal forcing (overwash, fringe) or that have a fast water turnover time (riverine) (Pool et al., 1975), as compared to land-locked systems (hammock) that have less hydraulic energy. Freshwater P influxes are high into riverine, moderate into basin and fringe forests and low into the scrub type (Day et al., 1989). Sheltered and land-locked systems are usually climate and/or groundwater dominated and behave as sinks of sediments and nutrients (Twilley, 1985). River and river-wave dominated systems are the two most important in terms of P transfer from the continents into coastal waters.
MANGROVE BIOMASS AND P CONTENTS
Values for biomass in various types of mangrove forests have been summarised by Lugo and Snedaker (1974) and Lugo et al., (1988). In most cases root mass was not determined. Few publications (Golley et al., 1975; Gong and Ong, 1990; Silva, 1992) indicate P concentrations in plant parts of mangrove forests (Table 2). The forests of Malaysia, Brazil, Panama and Puerto Rico have similar relative distributions for the various plant parts, although total biomass differs (Table 2). Most of the above-ground biomass in Rhizophora is concentrated in the trunks and in the aerial prop roots that provide mechanical support.
The P concentrations reported for woody and prop roots for the southeastern Brazilian mangrove are markedly low. The average whole plant-P concentration in the Malaysian and Panamanean mangroves are 0.91 and 0.80 mg g-1, respectively, and only 0.15 mg g-1 for the Brazilian forest. For comparison, a wet tropical primary forest in Manaus (Brazil) had a total biomass (above and below ground) of 43.8 g m-2 with a total P content of 6.4 g m-2 (Klinge et al., 1975), which yields an average concentration of 0.15 mg g-1, similar to that of the mangrove studied by Silva (1992). More data of P concentrations in the woody parts are necessary to understand the reasons for such wide variations.
The two very high dry-matter values shown for below-ground roots in Panama and Puerto Rico are due to the presence of peat (or fibrous mud) and to an unknown proportion of dead roots (Golley et al., 1962). Various studies have been done in the contrasting rhizosphere environments under Rhizophora, with a very high fibrous content, and the non-fibrous mud under Avicennia mangrove plants (Hesse 1961, 1963, Giglione and Thornton, 1965). Below-ground roots are very unevenly distributed in mangroves (Komiyama et al., 1987), contributing to the variability of the few available results, mainly when estimates are done by trench/pit methods (Gong and Ong, 1990). Up to 3 fold discrepancies have been reported between pit and allometry measurements of below-ground root biomass for mangroves (Gong and Ong, 1990).
PHOSPHORUS IN MANGROVE ESTUARIES
Results summarised by Alongi et al. (l992) show that SRP concentrations in unpolluted mangrove waters can vary between < 0.1 to 20 µM, while for DOP they can vary from undetectable levels to about 3 µM. Local characteristics as well as intertidal position, season and time of sampling can account for the wide range of concentrations. Variations due to tide and site of sampling can be illustrated by data from Ovalle et al. (l990): at the mouth of the mangrove creek in Sepetiba Bay, Brazil, SRP concentrations changed from 0.74 to 0.53 µM between low and high tide, respectively. At the upstream end, these concentrations were 1.16 and 0.53, respectively.
Table 2. Total biomass, P concentrations and P content in various parts of Rhizophora plants.
1 Managed forest, Malaysia (Gong and Ong, 1990); data were multiplied by 3.3, as suggested by the authors, since the corrected values where closer to those by Golley et al. (1975) for all plant parts, and to those by Silva (1992) for leaves and below-ground roots. 2 Brazil, Silva (1992). 3 Panama, Golley et al. (1975); does not distinguish between trunk and branches, so we attributed the dry mater mass to trunks. 4 Puerto Rico, Golley et al. (1962).
Plant parts ref. Biomass Dry Weight P conc. P content kg m-2 % mg g-1 g m-2 Leaves 1 1.72 8.5 0.66 1.14 2 0.40 5.5 0.70 0.28 3 0.35 1.0 0.85 0.30 4 0.78 6.9 - - Twigs/branches 1 2.43 12.0 0.99 2.41 2 1.29 17.6 0.001 0.001 4 1.27 11.2 - - Trunk 1 11.14 55.0 0.99 11.03 2 3.14 42.9 0.02 0.7 3 15.90 45.5 0.90 14.31 4 2.80 24.8 - - Prop roots 1 3.44 16.0 0.66 2.27 2 1.68 22.9 0.02 0.03 3 11.60 33.2 0.70 8.12 4 1.44 12.8 - - Root (below-ground) 1 1.72 8.5 0.99 1.70 2 0.82 11.1 0.90 0.74 3 7.40 21.2 0.70 5.18 4 5.00 44.3 - - Total 1 20.45 100 0.91 18.55 2 7.33 100 0.15 1.12 3 34.90 100 0.80 27.91 4 11.29 100 - -
Total P concentrations in mangrove soils (mainly Entisols, some Histosols, Lacerda et al., 1993) also vary (100-1600 mg kg-1), particularly as a function of sediment source, whether this is of continental or marine origin. Sediments in a large deltaic forest are richer in silt and clay, which reflects positively in their nutrient content. In riverine forests, sediments have a greater fine sand content and are nutrient-poorer (Boto and Wellington, l984).
Published results for total P concentrations, their depth distributions or the chemical forms of P in mangrove soils are scarce (Boto and Wellington, l984; Silva, l992), as opposed to other estuarine or marine environments (e.g. Yamada and Kayama, 1987; Balzer, 1986; Salomons and Gerritse, 1981). Hesse's (1963) data for Sierra Leone show that total P in the first few cm of soils, 700 mg kg-1, decreases sharply to about 350 mg kg-1 at a depth of 10 cm and remains almost unchanged down to 40 cm, from where it increases gradually to 500 mg kg-1, at 1.5 m depth. Most of the P depletion in the 10 to 30 cm layer is associated with the fibrous mud found under Rhizophora trees, and shows the influence of plant uptake. About 50% of the total P content in the surface layer is organic P, increasing to 85% (280 mg kg-1) in the fibrous mud. Based on the 119 mg g-1 organic C content found at this depth (Hesse, 1961), the organic C:organic P weight ratio is 425, which indicates favourable conditions for P immobilisation. The C:N ratio of this fibrous mud was 36, also indicating a strong N immobilisation potential, experimentally confirmed through additions of ammonium sulfate (Hesse, l961).
Silva (1992) found a much lower average total P content, 180 mg kg-1, in the 0-25 cm layer of a Rhizophora mangrove forest in SE Brazil, of which 55% was Ca-P and the remainder was equally divided between organic P and Fe-Al-P. These results do not reproduce the patterns of P distribution with depth obtained by Hesse (1963). While this last author sampled the soil under the Rhizophora trees, Silva (1992) appears to have sampled between trees, although this is not clearly stated in his methodology. Total P content showed large and irregular fluctuations within the 25 cm depth (Silva, 1992), possibly due to changes in the particle-size distribution of the sediments. In four 6 cm-layers of mangrove sediments (0-24 cm), silt decreased from 38% to 1% while clay increased from 55% to 88%, an sand varied between 48% to 69% (Giglioli and Thornton, 1965). Since P is associated with the finer fractions (Salomons and Gerritze, 1981) these fluctuations in particle-size distributions will certainly affect P concentrations too.
Phosphate concentrations in pore water have been found to be higher than in overlying water. At 25 cm depth, SRP in the interstitial water varied between 3-51 µM (Ovalle et al., l990; Silva, 1992). However, a detailed profile in a Missionary Bay sediment showed that concentrations of SRP down to 15 cm were low (<2 µM) followed by a large increase in the next depth increment (Alongi et al., l992). It is possible that this increase is related to the beginning of the anoxic layer.
Redox potentials (Eh) in alternating aerobic/anaerobic conditions in sediments can control the P solubility, through changes in the Fe(III)/Fe(II) ratios. There are few data of Eh and its variability with depth for mangrove soils. Negative Eh values determined by Boto and Wellington (1983) indicated reducing conditions (< +100 mV indicates absence of oxygen, Day et al., 1989), but there was no indication of sampling depth and tidal levels. Interstitial water sampled at a 30 cm depth in the Sepetiba (Brazil) mangrove forest (Ovalle et al., 1990) had strongly reducing conditions (Eh from -110 to -362 mV), while creek water had Eh values between 122 and 282 mV and O2 concentrations between 3.6 and 6.5 µM. As in lake and marine sediments (Salomons and Gerritze, 1981; Gächter et al., 1988), overlying waters with these O2 concentrations will maintain oxic conditions at least in the first few millimeters of sediment at high tide. During low tide, mangrove sediments will likely have oxic conditions to greater depths, as the moisture content decreases from 60% (field moisture at low tide) to 40-30%, the onset of oxic conditions (Hesse, 1961). The water holding capacity at which oxic conditions are established will differ between sites and layers depending on particle-size distributions.
Except for Hesse's preliminary work, the effects of an upper oxic layer and a lower, permanently anoxic one, on solubility and diffusion of P in sediments has not been studied in detail in mangrove systems. In addition to these two layers, mangrove systems, as salt marshes (Day et al., 1989), have an additional oxic environment in the vicinity of the root system. It is accepted that low redox potentials in anoxic layers promote the formation of Fe (II) through dissolution and reduction of Fe (III) oxi-hydroxides with release of associated P (e.g. Sundby et al., 1992). This promotes an increase in the SRP concentration in interstitial water, with Fe being converted to insoluble iron sulfides (Rosenfeld, l979; Krom and Berner, 1980; Caraco et al., 1989). This process is catalyzed by enzymes released by nitrate- and sulfate-reducing bacteria and depends on a carbon source (Jansson, 1987). When the redox potential increases, sulfides are oxidised to sulfate and Fe(II) to Fe(III). The increase in Fe (III) promotes the precipitation of oxi-hydroxides which re-adsorb PO4. The half-time for this adsorption process in seawater at pH=8.0 is 1 min (Crosby et al., 1984). A conceptual model for the transfer of P across the oxic-anoxic boundary layer and the mechanisms involved has been proposed by Sundby et al. (1992).
There is alternative evidence indicating that the related changes in Fe and P concentrations do not necessarily prove that the cycling of these two elements is coupled (Gächter et al., 1988). In a very detailed study in Lake Sempach, these authors followed changes in pH, and concentrations of O2, total and particulate Fe, C, P and Mn, along various cycles of oxic and anoxic conditions. Their results emphasise the role of bacteria and other microorganisms as sinks and sources of P, in conditions similar to those that promote the decrease and increase of Fe concentrations, respectively. In addition, between 22 and 80% of the total particulate P content in a core taken from the same lake, was in the bacterial biovolume. Furthermore, unsterilised sediments with no C and N additions could sorb up to 31% more PO4 than could sterilised sediments. This last effect was also noted in the isotopic exchange studies of 32P by Pomeroy et al. (1965).
Additional evidence of the importance of microbial P cycling in mangrove forests is the flux of 200 g C m-2 y-1 into the bacterial biomass (Robertson et al., 1992). This is 10% or more of the net primary productivity of mangroves (Table 3). Weight C:P ratios in bacteria can vary considerably (Bratbak, 1985; Stewart et al., 1987). Using an average value of 25:1, the C flux would be associated with a P flux of 8 g m-2 y-1, which exceeds by a factor of 2-3 the needs of the mangrove trees (Table 3). This probably means that the microbial biomass strongly competes with plants for P, as has been demonstrated in the C-rich environment of the litter mat of humid tropical forests (Salcedo et al., 1991b).
PRODUCTIVITY OF MANGROVE FORESTS
The data for litterfall production in several types of mangrove forests, some of them with mixed stands of Rhizophora and Avicennia, are shown in Table 3. From data compiled by Twilley et al. (1986) and Day et al. (1987) litterfall follows the order riverine > basin > fringe > scrub. However, data by Boto and Bunt (1982) for Australia and by Silva (1988) for Brazil show that fringe mangroves can have much larger litterfall values than previously reported. To estimate the amount of P transferred by average litterfall for each mangrove type, we used the average P content (0.28 mg g-1 in DM) found by Boto and Bunt (1982). The resulting estimates for P flux are of the same order of magnitude of litterfall-P measured in an undisturbed primary wet tropical forest in the Atlantic coast of northeastern Brazil (0.34 g P m-2 y-1, Sampaio et al., 1988).
Table 3. Litterfall, net primary productivity and estimated P content in litterfall, and P requirements for riverine, basin , fringe and scrub mangrove forests.
Riverine Basin Fringe Scrub Litter fall g C m2 y-1 5841 2971 4051 841
5632 3762 9493 8914 g P m-2 y-1 0.365 0.195 0.415 0.055 Net PP g C m2 y-1 11062 9206 7232 1717 20696 g P m2 y-1 2.46-4.608 2.048 0.608 0.388 1 Twilley et al. (1986), several authors. Calculated assuming 450 mg C g-1 DM. 2 Day et al. (1987) also assuming 450 mg C g-1 DM. 3 Silva (1988). 4 Boto and Bunt (1982) 5 Estimated using a P concentration in DM of 0.28 mg g-1 (see text) 6 Day et al. (1989) 7 Teas (1979) 8 estimated using 1 mg P g-1 DM, for new material (see text).
Mangrove ecosystems typically exhibit among the highest production rates of any aquatic ecosystem, when daily C fixation rates of 1 to 5 g C m-2 are extrapolated for a year (Kjerfve and Lacerda, 1993). The value from Day et al. (1989) for a riverine mangrove in Puerto Rico (2069 g C m-2 y-1, Table 3) compares with the biomass accumulation of 45 t ha-1 y-1 measured in areas of high rainfall and moderate salinity, in the Indo-Pacific (Clough, 1992). Primary productivity in mangroves is often estimated through measures of litterfall, using a conversion factor of two (Day et al., 1987) (Table 3), although there is no clear evidence that they are related (Clough, 1992). To estimate the P required to sustain the productivities reported in Table 3, we used a P concentration of 1 mg g-1 for new fresh material and a 450 mg g-1 C content in the dry matter (Boto and Bunt, l982). Fringe, basin and riverine mangroves with similar productivities would need an annual P supply of about 20 kg ha-1 y-1, which is several times higher than the requirements of most agricultural crops.
Some of this P is probably derived from internal retranslocation, as suggested by the difference in P concentrations between litterfall (0.28 mg g-1) and new growth (1.0 mg g-1). Inputs through fresh sediments are an additional source of P (Hesse, 1963). The high P contents measured in the upper 3-5 cm of sediment, in contrast to the low concentrations found in the rooting zone, suggest that inputs from newly deposited material are important in maintaining the forest. High densities of mangrove roots and pneumatophores slow water circulation and favour sedimentation (Wolanski et al., 1992). Fine roots act as sediment binders (Hesse, 1961), in the same way as in terrestrial forests. Lynch et al. (1989) have determined sedimentation rates of up to 3 mm y-1 in Mexican mangroves. If we assume a P content of 600 mg kg-1 (19.4 µmol g-1) of sediment and a bulk density of 1.2 Mg m-3, this represents a P input of approximately 21 kg ha-1 y-1. However, erosion processes by cliffing, sheet wash and tidal creek extension are also operative (Woodroffe, 1992). An additional P supply is provided by the recycling of P in the litterfall.
PHOSPHORUS EXPORTS AND IMPORTS FROM MANGROVE SYSTEMS
It is difficult to quantify P fluxes between mangrove systems and coastal waters because this requires the knowledge of the contribution from various components with different P contents and residence times in the estuary. The exchange between mangroves and nearshore areas also needs to be considered. This exchange depends on the size of the system, nearshore circulation pattern and on the topography and coastal features such as reefs, sand banks, barriers, etc. Certain combinations of these variables could result in the transport of outwelled P to deep waters or trapping near the estuary mouth, from where it can be redistributed along the coast (Wolanski and Ridd, 1986) and/or return to the estuary (Medeiros and Kjerfve, 1993).
There are no data available yet integrating all these components into a P budget within a mangrove ecosystem. However, limited data have been published that give or allow the computation of partial fluxes, based on concentrations of various P forms, in addition to litterfall and litter being transported through the systems (Table 4).
The ratio between the mangrove swamp area and that of the open creek is high for the three fringe systems (Table 4). The presence of extensive mangrove swamps increases the time lag of high water between the head and mouth of the estuary and, thus, the water slope between these two points. This results in peak ebb currents often 20 to 50% (160% for the Klong Ngao) higher than the peak flood, which are not seen in estuaries without mangroves or having a small swamp:creek area ratio (Wolanski et al., 1980). Larger ebb currents favour exports of suspended P from tidal creeks.
Table 4. Average exchange of litter and P (g ha-1 day-1) between mangrove and coastal waters. DIP Dissolved inorganic P; TDP total dissolved P; TPP total particulate P. (-) net exports and (+) net imports.
1 Kjerfve and Wattayakorn (1990), 1a Wolanski et al. (1992), 1b Wattayakorn et al. (1990), 2 Silva (1988; 1992), 2a Rezende et al. (1990), 2b Ovalle et al. (1990), 3 Boto and Bunt (1982), 3a Wolanski et al. (1992), 3b Robertson (1986), 3c Boto and Wellington (1988), 4 Gong and Ong (1990),
Klong Ngao1 Sepetiba2 Coral Cr.3 Matang4 (Thailand) (Brazil) (Australia) (Malaysia) Area (ha) 200 42a 500 40,800 Type Fringe Fringe Fringe Riverine Tides sampled 1-8 5 14 Swamp:creek 2.71a 21.4* 5.53a Litter fall 2,352 (neap)1b 31,6592b 24,400 26,648 51,744 (spring)1b Litter export -15,3003b -23,000 leaves -6 to -2501b -10,600 twigs+fruits -12,400 DIP +43 to -583 +54 to 0 3.63c +6.3 to -1.92b TDP -13001b +0.05 to -94 +13.73c TPP** +0.14 to -326 -6.8 -29.7 *Calculated from ref. indicated on the top of column.
**Includes small litter.
The Klong Ngao mangrove system experiences macrotides (spring range 4.4 m) and is completely inundated once a month, which facilitates nutrient exchange between mangrove swamps and creeks (Wattayakorn et al., 1990). Net export of leaf litter varies from 6 to 250 g ha-1 day-1 during the wet season, which is very low compared to Matang forest but of the same order of magnitude as values for a Phuket mangrove system (900 g ha-1 day-1) during spring tides (Poovachiranon and Chasang, 1982). Phosphorus fluxes were computed based on current-concentration data (Kjerfve, 1990) and also using a mathematical model (Wattayakorn et al., 1990). The modelled results were considered by the authors as more reliable, due to uncertainties in current data. Computation of P fluxes indicate an outwelling of total dissolved P of 1,300 g ha-1 day-1 (Table 4).
While the P content of suspended sediments was not measured, an inflow of turbid waters from the Kro Buri river into the Klong Ngao was observed. The much smaller residence time of the turbid water (0.5 days) relative to the flushing time for the system (6.5 days) indicated that fine sediments and associated adsorbed P load, was being trapped in the mangrove swamp. During the wet season, nutrient inputs from runoff could only account for one-tenth to one-fifth of nutrient exports (Wattayakorn et al., 1990). These authors also indicated the need to extend outwelling studies to the same time scale needed for the recovery of the estuary from a flood. This would reduce errors due to unsteadiness of freshwater inflows.
The mangrove forest at Sepetiba Bay is tidal-dominated and receives freshwater inputs through groundwater (a source of P and Si) and rainfall runoff (Ovalle et al., 1990). Tidal exchange is often the main mechanism operative in short creeks receiving low freshwater inputs, where the tidal prism represents a larger portion of the total water volume in the system.
Results by Silva (1992) and Ovalle et al. (1990) indicate that net imports of dissolved inorganic P are larger than net exports. Exports prevail during the rainy season, and imports during the dry season (Ovalle et al., 1990). At the same time, export fluxes of P as TDP are larger than imports (Silva, 1992), which indicates that the mangrove is contributing organic P into the coastal waters. Phosphorus is being exported mainly associated to the suspended particles, although some tidal cycles showed net imports (Table 4).
Sepetiba Bay is protected from wind waves and longshore currents by the Marambaia shoals. This tends to increase the residence time of outwelled waters in the bay. Thus, a large fraction of the water leaving during ebb tides, can re-enter the system in the next flood tide. Rezende et al. (1990) using C13/C12 ratios, verified that at all times, there was a landward contribution of marine particulate organic C (POC) and a seaward contribution from mangrove POC. Contributions from the mangrove into coastal waters were higher during spring and ebbing tides. Mangrove POC accounted for 65% and 21% of total POC during a spring ebb and flooding tides, respectively.
Coral Creek is also a tidal dominated mangrove system (3 m range) but without any influence from terrestrial runoff or groundwater. Leaves falling on lower intertidal banks are washed out completely at each tide and do not return to the system, accounting for a seaward net transport of 6 g ha-1day-1 (Boto and Bunt, 1982). Leaves falling on upper banks, flooded only at spring tides, stay longer in the system. About 17% of the total leaf fall is grazed/buried by sesarmid crabs but part of this P is later recycled as crab excretes (Robertson, 1986). Particulate organic matter (mostly associated to suspended sediments) contributes to an additional net P export of 0.08 g ha-1 day-1. On the other hand, computations of the exchange of total dissolved P revealed a net import of 13.7 g P ha-1 day-1 (Boto and Wellington, 1988), twice as much as exported through small litter.
Data for the Matang mangrove forest in Malaysia are based on measurements of litterfall, separated into leaves, twigs and fruits, and corresponding P concentrations obtained by Gong and Ong (1990) (Table 4). The authors estimate that 52% of the leaves, 50% of the twigs and 0% of the fruits are exported from the system, which yields a P flux of 29.7 g ha-1 day-1 (Table 4). However, this mangrove is of the riverine type, while the estimates were derived from fringe mangroves, which is probably not a valid extrapolation. Thus, until a better knowledge of the type of exchange with coastal waters as well as of the inputs through its watershed is obtained, no conclusions can be drawn as to the net balance of this system and of its contribution to coastal productivity.
In addition to inflows and outflows of P, much information is needed on biological recycling within the system (Peterson et al., 1988). The low hydrodynamic energy in floodplains allows for an intense P recycling through a complex foodweb. An example of such recycling is given for lake Calado, an inland floodplain in the Amazon basin, normally controlled by alternating seasons of high and low waters (Fisher et al., 1991). These authors showed that P inputs into the system originated from Amazon River waters (51%), groundwater (19%), adjacent lakes (13%) surface runoff from surrounding uplands (11%) and rainwater (6%). Averaged over a lake area of 5.85 km2, P inputs amounted to 45 µmol m-2 day-1 while outputs were 110 µmol m-2 day-1 (64% lake outflow, 31% burial and 5% groundwater). Thus, the system acts as a net P exporter. However, due to water column and sediment recycling, the small amount of P retained was able to sustain a complex food web, equivalent to an input of 4410 µmol P m-2 day-1 (1.37 kg ha-1 day-1).
Probably no such intense recycling can be expected in the much shorter times available for tidal dominated systems. In this respect, it is important to consider the various time scales of processes taking place within the system. Decomposition rates of materials vary from a few days (leaves) to two years (trunks). Tides can impose great variability, as gross transfers of material moved in and out of the estuaries are large, in contrast to small net fluxes. Thus, fortnightly and daily cycles need to be considered during samplings. While some variables appear to be unaffected by the tides, such as concentrations of dissolved constituents (Boto and Wellington, 1988), total P transport it is most likely to be affected by variation in tidal currents. In tropical areas seasonality of winds and rainfall has also to be taken into account. Local and coastal topography and morphology impose further constraints. Presence of coastal saline water masses can create a dynamic barrier, reducing the exchange of material between estuarine and coastal waters (Wolanski et al., 1980; Medeiros and Kjerfve, 1993). Due to the large number of variables involved, nutrient balances are very site specific for each system. More data are needed before basic generalisations can be made.
Phosphorus in the Global Environment.
Edited by H. Tiessen
© 1995 SCOPE. Published in 1995 by John Wiley & Sons Ltd.
Back to table of contentsLast updated: 12.07.2001